Explore how vSAN can be used to leverage local storage devices to create a shared datastore. Learn key concepts such as vSAN clusters, disk groups, cache and capacity tiers, and hybrid vs. all-flash architectures.
- [Instructor] In this video, we'll learn about VMware's Virtual SAN, or VSAN product, that leverages the local storage of ESXi hosts to present high performance shared storage across a cluster. I know that's a mouthful, so let's start with the very basics. With Virtual SAN, we are going to allow local storage to be used as shared storage. And this is really important, because for features like vMotion, or High Availability, or DRS, we have to have shared storage. We require shared storage for any of those features, and so if we're going to vMotion a VM from one host to another, if we're going to enable HA across a cluster, and allow virtual machines to fail over across that cluster, if a single host fails, or if multiple hosts fails, we have to have shared storage in order to enable those features. And same thing with DRS. If we're going to enable Distributed Resource Scheduler, we have to use vMotion for that, and therefore we require shared storage. So if you're going to set up High Availability or DRS, you need to create a cluster. Same thing with VSAN. It's created on a cluster of ESXi hosts, and it can leverage the local storage of those ESXi hosts in the form of what we call disk groups. Disk groups are local storage that is combined to form one large datastore that's accessible to all the hosts in the cluster. And that datastore is going to have SSD caching capability. So read-cache and a write-buffer built right into it to give high performance. Alright so let's start out with the very basics here. VSAN needs to be enabled on an ESXi host cluster. You can create clusters for a number of different purposes. It's essentially just a logical grouping of ESXi hosts. So for example, let's say you have these four ESXi hosts you see here, and a VM is running on host one. Well if we go ahead and we group those hosts available as a cluster, we can enable high availability. And if one of those hosts fails, the VM will reboot on some other host. Or we can enable DRS to have virtual machines automatically load-balance across those hosts. Or we can enable Virtual SAN, VSAN. So these are all different features that are available only on a cluster of ESXi hosts. So the storage on our ESXi hosts is going to be organized in something called disk groups. That's the logical unit of storage that Virtual SAN is going to essentially consume and share out across the cluster. So here on ESXi01 you can see we've got two disk groups. Each disk group in this case has 6 hard disk drives an one solid state drive. You have a host, 02, has one disk group. Host 03 has one disk group. And you can set this up to automatically happen. So it's called Disk Claim Mode in Virtual SAN, and you can set up what's called Automatic Disk Claim Mode. So what that means it that as you install SSD and hard disk drives onto your ESXi hosts, they will be automatically claimed and allocated to VSAN disk groups. Or you can use what's called manual claim mode, and in manual claim mode, you will manually create the disk groups and choose which storage devices go in which disk group. There's really only a few rules that you have to bear in mind. So for each disk group, you have to have at least one SSD device. And then you can either have a group of one or more magnetic devices like hard disk drives, and those are called our capacity devices. So the purpose of the SSD is purely to be used as a read-cache and a write-buffer. The purpose of the capacity devices, which in this case are hard disk drives, is the persistent storage of data. The long-term persistent storage of data. And these capacity devices, we have a couple of options here. All right, in a hybrid configuration, the capacity devices are going to be hard disk drives. That's our hybrid configuration. Right, it's part SSD, part hard disk. Or we could go with an all flash configuration in which case our cache device will be SSD. And all of these capacity devices will also be SSD. And I know my highlighter doesn't really make a great pen here, but that says SSD. (laughs) So that's the all-flash configuration, which obviously is going to out perform the hybrid configuration. So if you need extremely fast all-flash performance, then you should go with an all-flash VSAN configuration. And there's features specifically that make the all-flash version a little bit more cost-effective. Because we all know SSD storage, really expensive, right? The price you're going to pay per gigabyte is much higher than you would pay with traditional hard disk drives. So going with an all-flash configuration can get really expensive really quickly. Now with that being said, the current version of VSAN supports features like RAID 6 and RAID 5, erasure coding, and other space-reclamation techniques specifically on an all-flash VSAN configuration. Those features are not supported on the hybrid. They're only supported on an all-flash. So if you're kind of thinking, well which way should I go, should I go all flash? Should I go hybrid? Bear that in mind, right? That if you take advantage of some of the space-saving techniques that are introduced in the current version of VSAN, you can potentially get that all-flash configuration to approximately somewhere near the price of your hybrid deployment, because of the fact that you can reduce the amount of data that you need to store. So I know I kind of diverged off on a bit of side topic there, (laughs) but back to the very basics, right? So once we've created our ESXi host cluster, we can enable Virtual SAN on it. We can enable VSAN. And at that point, all of those disks in our disk group, they'll essentially be combined, to form one big datastore called our VSAN Datastore. VSAN Datastore will be comprised of all of the space, provided by the disk groups on all of the ESXi hosts. I mean the hosts can be part of the cluster, even if they don't actually contribute any storage. Which is kind of funny to think about, but you could potentially add a fourth ESXi host with no local storage, and let it just use the capacity of the VSAN Datastore. But with that being said, you have to have storage on the first three hosts in the cluster. And most of the time it's not really recommended to have a host with no local capacity as part of the VSAN cluster. Okay so the basic concepts of VSAN are what we've covered so far. We know that, number one, we're going to leverage local storage to create a shared datastore. We know that that local storage can either be 100% SSD, or it can be a hybrid configuration, where it's got an SSD read-cache and write-buffer. But on the backend, all of the capacity devices are traditional, physical hard disks.
- Storage commands
- Virtualized operating systems
- Securing storage traffic
- Network topology
- Connections and arrays
- VMkernel port
- Local and shared storage
- Storage adapters
- Configuring encryption options
- Creating clones and snapshots
- Storage performance and availability
- Failure management
- Analyzing metadata